ChatGPT has hit roughly 1 billion weekly active users, a milestone that would make it one of the most-used software products in history. At the same time, OpenAI dropped GPT-5.6 Luna pricing to $0.20 per million input tokens (an 80% cut) putting frontier AI within reach of solo developers and small teams who could not afford it before.
Why it matters: When the price of intelligence drops this fast, the competitive advantage shifts from access to what you build with it. If your plan is to "wait until it's affordable," that moment has arrived.
EU officially bans AI-generated sexual deepfakes, and labels all AI content
The European Union struck a legislative deal explicitly banning nudifier apps and AI systems that generate non-consensual sexualised imagery. Separately, from August 2 the EU AI Act now requires that all AI-generated content (images, audio, video, text) be clearly labelled so people know what was made by a machine. Both apply across all 27 member states.
Why it matters: Europe just moved from discussing AI harm to legislating against it at scale. Platforms operating in the EU must now comply or face penalties. Other governments, including in Asia, are watching this enforcement playbook closely.
India mandates AI education in every school from Grade 3: 247 million students
India's Ministry of Education confirmed that AI will become a mandatory subject from Class 3 upward starting the 2026-27 academic year, aligned with the National Education Policy 2020. The Bharat Bodhan AI Conclave 2026 concluded in New Delhi this week, focusing on how to actually deliver this at scale across 1.47 million schools and 10 million teachers. The implementation gap (digital infrastructure, teacher training, culturally relevant content) remains the real challenge.
Why it matters: This is the largest AI-in-education rollout attempted anywhere on earth. If it works, India builds a generational advantage. If the infrastructure gap is not closed, the policy stays on paper. Either way, AI literacy is now an official part of childhood in India.
Humanoid robots are now doing real warehouse jobs, by the thousands
Figure, Tesla (Optimus), and Boston Dynamics have collectively deployed thousands of general-purpose humanoid robots into commercial warehouses. These are not demos: they walk independently, identify irregular objects, and adapt to changing environments in real facilities. The shift from controlled pilots to operational deployment happened faster than most industry forecasters expected.
Why it matters: Physical AI (machines that reason and move in the real world) is crossing the line from research to commercial reality. Logistics and warehouse sectors are the first to feel the impact, but the pattern will repeat in other physical industries.
Google's Gemini agent now calls stores and completes purchases for you
Google rolled out a consumer AI agent inside Gemini that can call stores, check inventory in real time, and complete purchases on your behalf, no manual searching, no checkout flow. This moves AI from answering questions to actually doing errands, crossing a threshold that voice assistants have been promising since 2016.
Why it matters: The era of AI that chats is giving way to AI that acts. When your assistant can make a phone call and buy something, the relationship between people and software changes fundamentally, for consumers and for every business that currently depends on human-driven purchase decisions.
AI data centres are straining the power grid, demand could double by 2027
US data centre power demand is expected to climb from 31 gigawatts in 2025 to 41 GW in 2026 and 66 GW by 2027, according to Goldman Sachs Research. That 66 GW figure is roughly equivalent to the output of 66 large nuclear plants. The energy workforce needed to build and maintain this infrastructure is already in short supply, with the US power sector needing an estimated 510,000 additional workers by 2030.
Why it matters: AI's environmental footprint is no longer a talking point, it is showing up in electricity bills and grid planning decisions. Countries and companies that can secure clean, affordable power at scale will have a structural advantage in the AI economy.
India's AI classroom ambition vs. its infrastructure reality
India just committed to the most ambitious AI-in-education rollout in history. 247 million students. 1.47 million schools. 10 million teachers who need training. The policy is real. The digital infrastructure, reliable internet in rural schools, and culturally relevant AI tools, not yet at scale. This gap between mandate and delivery is the story to follow through 2026 and into 2027. If India closes it, the payoff is a generation of AI-literate citizens. If it doesn't, it becomes the most expensive good intention in education history.
GPT-5.5 thinks before it speaks + Google's new video and music models
OpenAI's GPT-5.5 introduced a native "System 2" architecture, it pauses on hard problems and maps a logical step-by-step tree before generating an answer, rather than replying instantly. Meanwhile, Google quietly updated its media suite: Veo 3.1 for video generation and Lyria 3.5 for music, with richer melodies and more expressive vocals, both built into Google Flow.